You Give Me the Chills

Implicit Suspicion Embodied as Physical Coldness

The first assessment a perceiver makes of other people is how socially “warm” or “cold” they are—it is a first-pass friend-or-foe assessment that gives the perceiver a sense of whether to be suspicious or not (Asch, 1946; Cuddy, Fiske, & Glick, 2008). People assess the trustworthiness of faces faster than they assess any other trait (i.e., after just 100 ms of exposure; Willis, 2006). However, it is unclear (a) what nonverbal cues perceivers attend to when making assessments of interpersonal warmth or coldness and (b) how perceivers react when those nonverbal cues give them reasons to be suspicious.

We suggest that an inappropriate amount of behavioral mimicry may be an implicit cue to suspicion because it signals that something is “off.” For instance, it may signal social coldness. Given the potential evolutionary significance of detecting possible threats in the social environment, suspicious nonverbal cues may lead people to experience a generalized, diffuse feeling of physical coldness because this feeling would be consistent with related psychological concepts of social threat (e.g., concepts associated with the idioms “icy stares” and “cold receptions”). Indeed, research on embodiment has suggested that people often construe their abstract, social world in concrete, physical terms (Schubert, 2005; Williams, Huang, & Bargh, 2009), such that activating physical coldness may activate social coldness, and vice versa (see also Williams & Bargh, 2008).

In other words, people may literally feel cold in response to social cues. People who are made to feel lonely or socially excluded, for instance, make lower room-temperature estimates than do people who are made to feel socially included (Zhong & Leonardelli, 2008), and briefly experiencing physical coldness significantly increases feelings of loneliness (Bargh & Shalev, 2012). The same principle applies to perceptions of social threat: Knight and Borden (1979) found that socially anxious participants who were assigned a public-speaking task subsequently experienced heightened vasoconstriction—a physiological response to threat that is associated with cold skin and goose bumps. Vasoconstriction also occurs when an interaction partner behaves in stereotype-inconsistent ways (Mendes, Blascovich, Hunter, Lickel, & Jost, 2007), which suggests that perceived violations of social standards for behavior can indeed trigger feelings of coldness.

This type of embodiment appears to be rooted in neurobiology. The gut feeling people get about the trustworthiness of their interaction partners is linked to signals from the insula—the region of the brain that regulates feelings of trust (Meyer-Lindenberg, 2008), physical warmth and coldness, and certain types of behavioral mimicry (Insel & Young, 2001; Lee, Josephs, Dolan, & Critchley, 2006). For example, Kang, Williams, Clark, Gray, and Bargh (2010) found that the same region of the insula (left anterior) was activated by the physical sensation of coldness and by social coldness (in the form of betrayals) in an economic trust game. Given that mimicry is a primitive social behavior that both humans and nonhuman primates engage in (Chartrand & van Baaren, 2009), there also may be primitive links between implicit standards for mimicry and tendencies toward suspicion. An overlap among the neural systems for mimicry, trust, and embodied feelings of coldness may facilitate the coactivation or spreading activation that links perceived violations of implicit mimicry standards to physical feelings of coldness.

Note that a basic tenet of our account is that perceivers’ embodied reactions may go beyond direct reflections of the metaphorical warmth or coldness of manifest social cues. Although direct exposure to cold social cues can certainly elicit sensations of physical coldness (Williams et al., 2009; Zhong & Leonardelli, 2008), we suggest that perceivers’ reactions to multiple, often compounding social cues is dynamic and context dependent—interactive as opposed to additive. Implicit standards for nonverbal behavior may change with the situation, which means that mimicry can be appropriate in one moment yet inappropriate in the next. And within a given context, situation-behavior mismatches may signal social coldness. Therefore, it may be more “chilling” to be mimicked than to not be mimicked during a socially cold interaction because any mimicry would violate implicit standards for that type of interaction. Not only may such inappropriate mimicry be intrusive and deplete self-regulatory resources (Dalton et al., 2010; Liu et al., 2011), but a person’s violation of standards for mimicry could signal that that person is also not going to follow other standards for behavior—a potentially broader threat that can be inferred only from evaluating the underlying behavior in the context of the overarching situation.

Another basic tenet of our account is that violations of implicit mimicry standards should moderate felt coldness, but adherence to those standards should not moderate felt warmth. People react to negative deviations, but they tend not to react much to behaviors that meet their standards (Skowronski & Carlston, 1987). It is easier for perceivers to confirm interaction partners’ negative traits and disconfirm their positive traits than to do the reverse (Tausch, Kenworthy, & Hewstone, 2007), a fact that suggests a perceptual bias toward suspicion. Consistent with the notion that moderate physical warmth is people’s default state, research has shown that experiencing physical warmth does not lower self-reported feelings of loneliness as much as experiencing physical cold increases them (Bargh & Shalev, 2012, Study 2). Thus, although by default, people may want to trust others, they nevertheless remain highly vigilant to cues suggesting that they cannot (von Hippel & Trivers, 2011).

In sum, individuals may feel physically cold when other people violate implicit standards for mimicry. We tested this general hypothesis in three studies. Specifically, we examined how differences in the explicit style of an interaction (Study 1), dispositional differences in the individual (Study 2), and racial differences between the individual and the interaction partner (Study 3) can all moderate implicit standards for mimicry and thus determine whether it is the presence or the absence of mimicry that perceivers find chilling.

Study 1: Interaction-Style Differences

In our initial study, we sought to test whether the presence or absence of mimicry by an interaction partner would lead individuals to report different levels of felt coldness. Any friendly interaction is likely to involve at least a small amount of mimicry (Lakin & Chartrand, 2003), so in an affiliative context, a complete lack of mimicry would violate implicit behavioral standards. However, we argue that what instills felt coldness is not always the absence of mimicry per se, but rather the deviation from behavioral standards given the overarching style of the ongoing interaction. Mimicry is generally linked to affiliative intent (Chartrand & van Baaren, 2009), and it occurs more often among peers than among nonpeers (Cheng & Chartrand, 2003). Therefore, mimicry should not occur in interactions that are more task oriented than affiliative, because such interactions may be too polite and formal to warrant affiliative behaviors such as mimicry. Differences in the explicit style of an interaction (affiliative vs. task oriented) could change perceivers’ implicit standards for mimicry—and violations of either type of standard could be chilling for perceivers even if the conscious elements of the interaction remain perfectly mundane and pleasant. Thus, if one’s interaction partner uses an affiliative interaction style, then a lack of mimicry might violate behavioral standards and thus be chilling; however, if the interaction partner is task oriented and expresses no affiliative intent, then the presence of mimicry might violate standards and elicit feelings of coldness.

Participants in Study 1 interacted with an experimenter who used either an affiliative or a task-oriented interaction style while either mimicking or not mimicking the participants’ nonverbal behaviors. We hypothesized that participants’ subsequent feelings of physical warmth or coldness would depend on the experimenter’s interaction style. A lack of mimicry would match a task-oriented interaction but would not match an affiliative interaction; we therefore hypothesized that participants would feel colder when they were not mimicked in an affiliative interaction or were mimicked in a task-oriented interaction than when they were mimicked in an affiliative interaction or were not mimicked in a task-oriented interaction.

Results and discussion

We tested felt coldness and warmth separately using a 2 (mimicry: mimicry vs. no mimicry) × 2 (interaction style: affiliative vs. task oriented) analysis of covariance, controlling for participants’ race. ³ Results indicated a crossover interaction for felt coldness, F(1, 35) = 8.86, p = .005, η _p ² = .20, but no effects for felt warmth (Fs < 1.0). As illustrated in Figure 1, participants in the affiliative condition reported feeling colder when they were not mimicked than when they were mimicked, F(1, 35) = 6.48, p = .015, whereas participants in the task-oriented condition showed the reverse pattern, F(1, 35) = 2.88, p = .098.

OPEN IN VIEWER

Fig. 1.

Results from Study 1: mean rating of felt coldness as a function of the experimenter’s interaction style (affiliative vs. task oriented) and presence or absence of mimicry. Participants rated how cold they felt using a scale from 1 (not at all) to 9 (extremely). Error bars represent standard errors.

Consistent with findings from past work on violations of implicit standards for mimicry (Dalton et al., 2010), our results revealed that participants who interacted with an affiliative experimenter felt colder if they were not mimicked than if they were. However, it was not the lack of mimicry per se that participants reacted to: The experimenter’s interaction style determined whether it was the presence or the absence of mimicry that participants found chilling. This result suggests that perceptions of an interaction partner’s use of mimicry depend on implicit standards associated with the context of the situation, and that perceivers’ embodied reactions are based on the combination of a given interaction partner’s level of mimicry and such situation-based standards for mimicry.

Study 2: Individual Differences in Self-Construal

In our second study, we examined how individual differences moderated participants’ embodied reactions to mimicry. Past work has shown that there are trait differences in how people construe themselves in relation to others—whether they regard themselves as socially interdependent or socially independent. Individuals with more interdependent self-construals prefer greater physical and psychological closeness with other people than do individuals with more independent self-construals (Holland, Roeder, van Baaren, Brandt, & Hannover, 2004; Oyserman, Coon, & Kemmelmeier, 2002). Although mimicry and feelings of interdependence are correlated (Ashton-James, van Baaren, Chartrand, Decety, & Karremans, 2007), mimicry can threaten one’s sense of independence (Liu et al., 2011). We argue that individuals with different self-construals differ in their standards for how much mimicry is appropriate in a given interaction, and that feelings of physical coldness result from deviations from these individual standards.

Participants in Study 2 engaged in a friendly interaction with an experimenter who either mimicked or did not mimic their nonverbal behaviors over the course of the interaction. Participants were then tasked with estimating the room temperature (a paradigm used in past research on embodied reactions to negative social feedback; Zhong & Leonardelli, 2008). We hypothesized that highly interdependent participants would make lower room-temperature estimates when they were not mimicked by the experimenter than when they were, and that highly independent participants would make lower room-temperature estimates when they were mimicked by the experimenter than when they were not.

Results and discussion

We used a regression analysis to predict participants’ temperature estimates from their mimicry condition (mimicry vs. no mimicry), their self-construal, and the interaction of these two variables, controlling for race. Results indicated a significant crossover interaction of mimicry condition and self-construal, b = 2.64, F(1, 30) = 5.55, p = .025, η _p ² = .17. Interdependent participants who were not mimicked reported marginally colder room temperatures than did those who were mimicked, b = 2.59, F(1, 30) = 3.34, p = .078. Conversely, independent participants who were mimicked reported marginally colder room temperatures than did those who were not mimicked, b = −2.69, F(1, 30) = 2.87, p = .100. The full pattern of results is illustrated in Figure 2. Note that given the constant room temperature of 72 °F, the observed differences in temperature estimates were errors of underestimation; estimates below 72 °F imply felt coldness resulting from violations of participants’ personal standards for mimicry.

OPEN IN VIEWER

Fig. 2.

Results from Study 2: perceived room temperature as a function of self-construal (independent vs. interdependent) and presence or absence of mimicry. Interdependent self-construal refers to a self-construal score 1 standard deviation above the mean; independent self-construal refers to a self-construal score 1 standard deviation below the mean. Error bars represent standard errors.

These results suggest that different perceivers react to mimicry in different ways. Whereas interdependent individuals felt colder when they were not mimicked, independent individuals felt colder when they were mimicked. These diverging sensitivities to mimicry highlight how individual differences moderate perceivers’ implicit assessments of social threat and thereby moderate the embodied reactions that follow.

Study 3: Racial Differences in Interaction Partners

We next considered how characteristics of one’s interaction partner can moderate embodied feelings of coldness in response to the presence or absence of mimicry. The nonverbal posturing and behavior used in interactions between people of different races can differ from the posturing and behavior used in interactions between people of the same race, often because of the negative arousal and anxiety inherent in cross- race interactions (Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997; Richeson & Trawalter, 2005). Recent work has suggested that mimicry may not be standard in cross-race interactions (Dalton et al., 2010); therefore, in Study 3, we tested the hypothesis that the absence of mimicry might be chilling in same-race interactions, whereas the presence of mimicry might be chilling in cross-race interactions.

Results and discussion

A 2 (mimicry: mimicry vs. no mimicry) × 2 (interaction type: same-race interaction vs. cross-race interaction) analysis of variance revealed a significant crossover interaction, F(1, 48) = 5.89, p = .019, η _p ² = .11. As illustrated in Figure 3, participants who engaged in cross-race interactions reported marginally lower room temperatures if they were mimicked than if they were not, F(1, 48) = 3.02, p = .089. Conversely, participants who engaged in same-race interactions reported marginally lower room temperatures if they were not mimicked than if they were, F(1, 48) = 2.90, p = .095. Among the participants who were mimicked, non-Caucasian participants reported lower room temperatures than Caucasian participants did, F(1, 48) = 4.28, p = .044. This pattern is consistent with results from our first two studies and with our theoretical model.

OPEN IN VIEWER

Fig. 3.

Results from Study 3: perceived temperature as a function of interaction type (same-race interaction vs. cross-race interaction) and presence or absence of mimicry. Error bars represent standard errors.

General Discussion

In three studies, we found that people literally feel colder in response to inappropriate amounts of behavioral mimicry and that an individual’s implicit standards for appropriate amounts of mimicry in a given interaction depend on the interaction style of the interaction partner (Study 1), the individual’s chronic disposition (Study 2), and racial differences between the two interaction partners (Study 3). A test of discriminant validity in Study 1 confirmed these increases in reported felt coldness, supporting the notion that suspicious nonverbal cues from interaction partners can elicit negative embodied reactions. The consistency of this effect across three different studies suggests that it may generalize to a range of social situations and relationships.

Our results suggest that people readily rely on implicit mimicry standards to assess and react to their social environment. Mimicry is typically considered an affiliative act (Lakin & Chartrand, 2003), so it is easy to see why mimicry that occurs in a nonaffiliative context might raise one’s suspicions. Indeed, in our studies, nonstandard amounts of mimicry seemed to evoke a threat response similar to that evoked by counterstereotypic behavior (Mendes et al., 2007). It would be interesting to examine whether such felt coldness corresponds with flight behaviors (e.g., behavioral distancing) or fight behaviors (e.g., behavioral backlash; Rudman & Fairchild, 2004), or whether it is merely conflated with psychological distance. These embodied reactions could also occur in lieu of behavioral responses, particularly in situations that do not warrant a behavioral response (Smith & Semin, 2007). Future studies could explore the behavioral correlates of these embodied reactions.

Our results also highlight the importance of (implicitly) knowing when to mimic other people and when it is okay to be mimicked by them, particularly if one wishes to minimize (or accurately assess) suspicion in zero-acquaintance contexts. This might require having properly calibrated standards for mimicry—that is, standards that are neither too broad nor too narrow. Having too broad a standard for how much mimicry is appropriate might increase one’s susceptibility to deception and threat, yet having a standard that is too narrow might facilitate unnecessary suspicion. Overly narrow standards may be especially problematic when one’s interaction partner’s mimicry is impaired because of developmental or physical disabilities or is guided by different cultural standards. An inability to establish trust at such an implicit level may undermine any subsequent efforts to get along.

It is still unknown how prevalent these embodied reactions are, and the answer might depend on the underlying cognitive mechanisms at work. In the present studies, participants’ standards for mimicry were presumably based on rather broad aspects of interactions (e.g., affiliative interactions, interdependent orientations), which suggests that the effects are generalizable. Nevertheless, it remains to be seen whether any kind of social mismatch is equally chilling. The observed effects could be limited to broad social norms that activate expectancies (whose violation leads to vasoconstriction) or to preferences (i.e., thwarted subgoals for how the interaction should unfold). Future studies could test for these boundary conditions, though we should note that the social-cognitive underpinnings of mimicry—including its interface with other mental processes—are still being explored, so there may be yet-unidentified factors that further moderate the prevalence of embodied reactions. For instance, if one’s interpersonal motives deviate from those typically involved in a particular type of interaction, then one may assess an interaction partner’s violations of standards for that type of interaction more positively, and they would thus be less chilling.

In conclusion, our findings suggest that people implicitly rely on mimicry cues to evaluate and react to other people. This means that everyday feelings of coldness may often stem from individuals’ implicit suspicions about an interaction partner’s nonverbal behavior. If an interaction partner’s mimicry can be assumed to be properly calibrated and thus meaningful, then it may be a cue to deception or threat. If this is indeed the case, then perhaps people are well served to beware of any strangers bearing mimicry.